fig = plt.figure(figsize =(14,14))
gs = fig.add_gridspec(10,10)
gs.update(wspace = 10,hspace = 2)
#ax0 = fig.add_subplot(gs[:,:])
ax1 = fig.add_subplot(gs[1:5, 0:5])
ax2 = fig.add_subplot(gs[1:5, 5:10])
ax3 = fig.add_subplot(gs[6:10, 2:8])
axes = [ax1,ax2,ax3]
data_ = df.copy()
## data with outlier removal
X_orig = data_.drop(columns = ['diagnosis']).values
X_perm = data_.drop(columns = ['diagnosis'])[perm_imp_feats1].values
y_orig = data_['diagnosis'].values
## data with VIF featue selection
X_vif = data_[vif_features].values
# plots
pca_viz(feat = X_perm,tar = y_orig,ax=axes[0]).visualize_data()
pca_viz(feat = X_vif, tar = y_orig,ax=axes[1]).visualize_data()
pca_viz(feat = X_orig,tar = y_orig,ax=axes[2]).visualize_data()
# text and title
## titles and text
axes[0].text(-50,570, 'Permutation Feature Selection',{'fontfamily':'serif','size':14, 'weight':'bold'}, zorder =3)
axes[1].text(-100,22.5, 'VIF Feature Selection',{'fontfamily':'serif','size':14, 'weight':'bold'})
axes[2].text(0,675, 'Original Features',{'fontfamily':'serif','size':14, 'weight':'bold'})
## add scores
#permutation annotations
axes[0].annotate('Acc: {}'.format(perm_list[0]),(1600,440),(1600,440), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
axes[0].annotate(' R2: {}'.format(perm_list[1]),(1600,400),(1600,400), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
axes[0].annotate('AUC: {}'.format(perm_list[2]),(1600,350),(1600,350), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
#VIF annotations
axes[1].annotate('Acc: {}'.format(vif_list[0]),(1000,18),(1000,18), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
axes[1].annotate(' R2: {}'.format(vif_list[1]),(1000,16.5),(1000,16.5), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
axes[1].annotate('AUC: {}'.format(vif_list[2]),(1000,15),(1000,15), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
# Original annotations
axes[2].annotate('Acc: {}'.format(orig_list[0]),(1900,510),(1900,510), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
axes[2].annotate(' R2: {}'.format(orig_list[1]),(1900,470),(1900,470), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
axes[2].annotate('AUC: {}'.format(orig_list[2]),(1900,425),(1900,425), zorder =3, annotation_clip = False,
fontsize=12,fontfamily='serif',fontweight ='bold',ha='center', color='black',alpha = 0.85)
# text and titles
fig.text(0.05,0.95,' Linear model Perforamace with PFI,VIF,and Original Data', {'fontfamily':'serif','size':20, 'weight':'bold'}, alpha = 1)
fig.text(0.05,0.9,'''Though the feature selection is done based on r2 metric, for comparision
of accuracy,r2, and auc scores among PFI, VIF and Orginal data with LinearRegression.
VIF based feature selection should give edge here... ''',{'fontfamily':'serif','size':14, 'weight':'normal'}, alpha = 0.9)
fig.text(0.40,0.88, "Cancerous",{'fontfamily':'serif','size':16, 'weight':'bold', 'color':colors[0]})
fig.text(0.52,0.88, '|',{'fontfamily':'serif','size':16, 'weight':'bold'})
fig.text(0.55,0.88, "Healthy",{'fontfamily':'serif','size':16, 'weight':'bold','color':colors[2]})
fig.text(0.75,0.075,'© Made by Milon',{'fontfamily':'serif', 'size':9,'weight':'bold'}, alpha = 0.8)
fig.show()
